Het effect van vraagtype in een internationale landenvergelijking
Er vindt geregeld grootschalig internationaal vergelijkend onderzoek plaats.
Denk aan PISA, TIMMS of PIRLS. Beleidsmakers gebruiken de resultaten uit deze
onderzoeken om beslissingen te nemen over het onderwijs. Hoewel de vergelijkingen
waardevol zijn, is er ook kritiek. Zo bestaan de toetsen bijvoorbeeld uit meerkeuzevragen,
terwijl dit vraagtype in sommige landen weinig gangbaar is. Wat doet dat
met de kwaliteit van de vergelijking? Kun je wel een eerlijke vergelijking maken als de toetsen in het ene land voor leerlingen toegankelijker zijn dan in het andere
land? In dit meerjarige onderzoeksproject belichten we deze vraag vanuit
verschillende perspectieven.
Making sense out of DIF
International educational surveys aim to evaluate and compare educational achievement across countries. For doing that, they design dozens of items, which are administered to thousands of students. Differential item functioning (DIF) analysis has been usually performed as a practical task to find items that, given the same level of individual proficiency, may present extra difficulty to students belonging to a particular group.
This project has been focused on promoting the idea of DIF as one of the most interesting outcomes in education, as it signals substantive differences and similarities among countries in terms of curriculum, educational practices, learning processes, etc. A variety of statistical and psychometric techniques have been investigated to find the ones that can be easily applied in practice and may contribute to investigate measurement non-invariance when there are many groups (e.g. countries).
This project has been focused on promoting the idea of DIF as one of the most interesting outcomes in education, as it signals substantive differences and similarities among countries in terms of curriculum, educational practices, learning processes, etc. A variety of statistical and psychometric techniques have been investigated to find the ones that can be easily applied in practice and may contribute to investigate measurement non-invariance when there are many groups (e.g. countries).
Facts
- Researchers: Edwin Cuellar, Ivailo Partchev
- Collaboration partners: University of Amsterdam, European Training Network OCCAM, Timo Bechger, Gunter Maris, Maarten Marsman
- Target group: Educational practitioners and policy-makers
- Duration: August 2018 – July 2021
"There is more to learn from the item-level performance than from a perfectly polished league table."